Senior Inference Optimization Engineer - Dragonfly Portfolio at Dragonfly

Department: Portfolio

Compensation

$180,000 – $250,000/yr

Setup: Hybrid
Type: Full-time
Level: senior
Posted: Aug 13, 2026

Summary from listing

Dragonfly is sourcing a Senior Inference Optimization Engineer for a portfolio company building privacy-first consumer AI infrastructure. The role focuses on improving LLM inference throughput, latency, and cost per token at scale, including GPU infrastructure, batching, attention, KV cache, speculative decoding, quantization, distributed inference, profiling, and load balancing. The position is remote in the United States, with compensation of $180,000–$250,000 for senior-level candidates.

Required Skills

GPU architecture
Parallel programming
vLLM
SGLang
Continuous batching
PagedAttention
KV cache
Speculative decoding
Quantization
CUDA graphs
Tensor parallelism
Nsight Systems
Python
Rust
Go
C++/CUDA
Triton kernels

Additional Information

Required Previous Experiences: 5+ years of performance optimization or HPC experience with GPU architecture and parallel programming
Hands-on experience with a production LLM inference engine at high volume
Experience optimizing LLM inference, including batching, attention, KV cache, speculative decoding, quantization, CUDA graphs, and torch.compile
Experience with distributed inference strategies across multi-GPU and multi-node environments
GPU profiling experience with Nsight Systems, Nsight Compute, and PyTorch Profiler
Proficiency in Python, Rust, or Go; C++/CUDA is a strong plus
Preferred Qualifications: Custom Triton kernels
Diffusion or image model inference optimization
Open-source inference framework contributions
Translation Company Name: Dragonfly